Sdarticle

Articles
Antiplatelet agents for prevention of pre-eclampsia:
a meta-analysis of individual patient data
Lisa M Askie, Lelia Duley, David J Henderson-Smart, Lesley A Stewart, on behalf of the PARIS Collaborative Group*
Background Pre-eclampsia is a major cause of mortality and morbidity during pregnancy and childbirth. Antiplatelet Lancet 2007; 369: 1791–98
agents, especially low-dose aspirin, might prevent or delay pre-eclampsia, and thereby improve outcome. Our aim Published Online May 17, 2007
was to assess the use of antiplatelet agents for the primary prevention of pre-eclampsia, and to explore which women DOI:10.1016/S0140-
are likely to benefi t most.
6736(07)60712-0
See Comment page 1765
Methods We did a meta-analysis of individual patient data from 32 217 women, and their 32 819 babies, recruited to *Members listed at end of article
31 randomised trials of pre-eclampsia primary prevention.
Centre for Perinatal Health
Services Research
(L M Askie PhD,
Findings For women assigned to receive antiplatelet agents rather than control, the relative risk of developing Prof D J Henderson-Smart PhD),
pre-eclampsia was 0·90 (95% CI 0·84–0·97), of delivering before 34 weeks was 0·90 (0·83–0·98), and of having a NHMRC Clinical Trials Centre
pregnancy with a serious adverse outcome was 0·90 (0·85–0·96). Antiplatelet agents had no signifi cant eff ect on the (L M Askie), University
risk of death of the fetus or baby, having a small for gestational age infant, or bleeding events for either the women or of Sydney, Sydney, Australia;
UK Cochrane Centre, Oxford,
their babies. No particular subgroup of women was substantially more or less likely to benefi t from antiplatelet agents UK (L M Askie); Nuffi eld
than any other.
Department of Medicine,
University of Oxford, Oxford,
Interpretation Antiplatelet agents during pregnancy are associated with moderate but consistent reductions in the UK (L Duley MD); and Centre for
Reviews and Dissemination,
relative risk of pre-eclampsia, of birth before 34 weeks’ gestation, and of having a pregnancy with a serious adverse University of York, UK
outcome.
(Prof L A Stewart PhD)Correspondence to: Introduction
pre-eclampsia has been dampened, because once again Dr Lisa M Askie, NHMRC Clinical Pre-eclampsia is a multisystem disorder of pregnancy the promising results of a small trial were not supported Trials Centre, University of that is usually associated with hypertension and by subsequent larger studies.18 Although results of further Camperdown, New South Wales, proteinuria. The condition complicates 2–8% of trials are awaited, it now seems unlikely that antioxidants 2050, Australia
pregnancies,1 and can lead to liver and renal problems, will off er major benefi t for women at risk of pre-eclampsia. [email protected]
convulsions (eclampsia), and abnormalities of the clotting Thus, better understanding of the eff ects of antiplatelet
system. Since the condition adversely aff ects the placenta, agents currently off ers the best potential for improving
risks for the baby include poor intrauterine growth and outcomes for women at risk of pre-eclampsia. The PARIS
premature birth. Worldwide, 10–15% of the half million (Perinatal Antiplatelet Review of International Studies)
maternal deaths that occur every year are associated with Collaboration was formed to do a systematic review and
hypertensive
mainly meta-analysis based on individual patient data to assess pre-eclampsia and eclampsia;2 99% of these occur in the use of antiplatelet agents for the primary prevention low-resource countries.3,4 of pre-eclampsia and to explore which women are most The cause of pre-eclampsia remains unclear. likely to benefi t from such treatment.19 Nevertheless, disordered trophoblast invasion of the
maternal spiral arteries in early pregnancy is known to Methods
lead to underperfusion of the placenta and, ultimately, Search strategy and selection criteria
placental ischaemia and infarction.5 The resultant placental We searched the comprehensive register of trials devel-
damage is thought to lead to activation of platelets and the oped and maintained by the Cochrane Pregnancy and
clotting system6,7 and to an imbalance between prostacyclin, Childbirth Review Group. Details of how this register is
a vasodilator, and thromboxane, a vasoconstrictor and maintained are available elsewhere,20 but it involves
stimulant of platelet aggregation.8,9 The hypothesis that extensive searching of bibliographic databases such as
antiplatelet agents might prevent or delay pre-eclampsia Medline, the database of randomised controlled trials in
has been widely tested in randomised trials. The optimism the Cochrane Library, and searching relevant journals by
following early trials was later dashed by the results of hand. PARIS trialists were also asked if they knew of any
larger studies.10–14 Although systematic reviews of aggregate further studies. The search was last updated in December,
data show modest reductions in the relative risk of 2005.
pre-eclampsia, preterm birth, and baby death associated
Studies were included if they randomised women at with antiplatelet agent use,15 controversy remains.16,17 risk of developing pre-eclampsia to receive one or more Recent enthusiasm that antioxidants—particularly the antiplatelet agents (eg, low-dose aspirin or dipyridamole) combination of vitamins C and E—might prevent versus a placebo or no antiplatelet agent. To reduce the www.thelancet.com Vol 369 May 26, 2007
Articles
possibility of bias, quasirandom study designs—eg, those failure, liver failure, haemolytic anaemia elevated liver
using alternate allocation—were excluded. Methods of enzymes low platelet count [HELLP] syndrome, stroke),
treatment assignment and allocation concealment were non-spontaneous labour (induced labour or pre-labour
confi rmed with the trialists. Trials that included women caesarean), caesarean delivery, post-partum haemorrhage
who started treatment post partum or had a diagnosis of (blood loss ≥500 mL if supplied or trialists’ defi nition),
pre-eclampsia at trial entry were excluded. Each infant admission to neonatal special care or intensive
potentially eligible study was assessed independently by care unit, ventilation required by neonate, and neonatal
at least two members of the steering group, unblinded to bleeding.
authorship. Any diff erences of opinion regarding the
assessment of the inclusion criteria were resolved by Statistical analysis
discussion.
Analyses included all women randomised and were based Primary prevention was defi ned as antiplatelet agent on intention to treat. Each analysis was restricted to those use for women deemed to be at risk of pre-eclampsia, trials that had at least 80% of data available for that gestational hypertension, or intra-uterine growth particular outcome. The main analysis used a two stage restriction based on either their previous pregnancy approach: outcomes were analysed in their original trial history, a pre-existing medical condition (eg, renal and then these individual results combined in a disease, diabetes, immune disorder, chronic hyper- meta-analysis to give an overall measure of eff ect. A fi xed tension), or obstetric risk factors early in their current eff ect model was used, and the level of heterogeneity pregnancy (eg, being a primigravida or a having multiple assessed with the I² statistic.21 Random eff ects models pregnancy). Trials that recruited women in both primary were also run to test the robustness of results to choice of and secondary prevention settings were divided in such a model. Numbers needed to treat or harm were calculated way that only women enrolled in a primary prevention based on control event rates in the included trials. Analyses setting were included in these analyses. were done with SCHARP software, version 4.0.
The main analyses compared the eff ect of antiplatelet Data collection
agents versus placebo, or no antiplatelet agent, for each Data to be collected were agreed after extensive consultation outcome. Subgroup and sensitivity analyses were within the PARIS Collaborative Group. Anonymised data restricted to the main outcomes. Extra maternal and for each of the pre-specifi ed variables were requested for infant outcomes were also assessed to examine potential each woman randomised. Data were supplied in a variety benefi ts and harms.
of formats, re-coded as necessary, and were checked for To explore the eff ects by trial-level characteristics we internal consistency, consistency with published reports, prespecifi ed analyses, based on aspirin-only trials, and for missing items. Information about the trials—eg, grouped by an intended daily dose of 75 mg or less, or randomisation method and antiplatelet dose—were more than 75 mg. Owing to small numbers, a planned cross-checked with published reports, trial protocols, and third group (≥150 mg) was not created and relevant trials data collection sheets. Quality and integrity of the were included in the more than 75 mg group. randomisation processes were assessed by reviewing the To explore the eff ects by participant-level characteristics chronological randomisation sequence and pattern of we prespecifi ed subgroups based on (1) risk factors at assignment, as well as the balance of baseline trial entry, including whether normotensive, previous characteristics across treatment groups (taking into hypertensive disorders of pregnancy, diabetes, renal account stratifi cation factors). Inconsistencies or missing disease, multiple pregnancy, maternal age, previous data were discussed with relevant trialists and corrected small for gestational age infant, parity, and by type of when necessary. Finalised data for each study were verifi ed hypertension at trial entry, and (2) gestation less than with the relevant trialists.
20 weeks or 20 weeks and greater at trial entry. Four main outcomes were prespecifi ed: pre-eclampsia Sensitivity analyses were done excluding studies (hypertension with new onset proteinuria at or beyond without a placebo and by including studies irrespective 20 weeks’ gestation); death in utero or death of the baby of whether data were available for less than 80% of before discharge from hospital; preterm birth at less than participants. Variations in the defi nition of pre-34 weeks’ gestation; infant small for gestational age at eclampsia22–24 were also explored. Planned analyses of birth (as defi ned by individual trialists); and pregnancy other quality measures—eg, adequacy of allocation with serious adverse outcome (pregnancy where the concealment and blinding—were not done because mother dies or develops pre-eclampsia or if any baby is almost all trials (26 of 31 trials, 99% of women) were of preterm, small for gestational age, or does not survive to good quality.
discharge from hospital). Prespecifi ed additional outcomes included: maternal Role of the funding source
death, ante-partum haemorrhage, placental abruption, The funding sources had no input into the study design, early onset proteinuria (before 34 weeks’ gestation), collection, analysis, or interpretation of the data, report See Online for webtable serious maternal morbidity (including eclampsia, renal preparation or in the decision to submit the paper for
www.thelancet.com Vol 369 May 26, 2007
Articles
publication. The corresponding author had full access to all the data in the study and had fi nal responsibility for “™Ž•‘†™Š‘Š™˜
”“™—”‘
š’‡Š—
Š‘†™Ž›Š—Ž˜
Š‘†™Ž›Š—Ž˜
the decision to submit for publication.
“Ǡ
“Ǡ
”‹™—Ž†‘˜
ǩȤȠʏǪ
 ǩȤȠʏǪ
țǧȤțǩțǧȣȟǢțǧȤȢǪ Š‘Ž›Š—žʙȞȟœŠŠ˜ǏŒŠ˜™†™Ž”“ țǧȤțǩțǧȣȞǢțǧȤȣǪ Š™†‘Ǡ‡†‡ž‰Š†™‡Š‹”—Š‰Ž˜ˆ†—ŒŠ ȟȣȟǠȜȠȟȜȝ țǧȤȜǩțǧȣȜǢȜǧțȞǪ 115 trials were identifi ed as potentially eligible for our ’†‘‘‹”—ŒŠ˜™†™Ž”“†‘†ŒŠŽ“‹†“™ țǧȤțǩțǧȣȜǢȜǧțȜǪ review. Of these, 50 were ineligible, for several reasons, —ŠŒ“†“ˆžœŽ™˜Š—Ž”š˜†‰›Š—˜Š țǧȤțǩțǧȣȠǢțǧȤȡǪ including an absence of a comparison group or because they recruited women with established pre-eclampsia †›”š—˜†“™Ž•‘†™Š‘Š™†ŒŠ“™˜ only. Two further trials were excluded from the analysis
after data collection because they were found to have Figure 1: Main outcomes for mother and baby
used quasirandom methods of treatment allocation. A *Pregnancy with any of four main outcomes above or maternal death. Fixed eff ect model used to calculate relative risks.
full list of ineligible trials is available on request. Thus,
63 trials (with 38 026 women) were eligible for inclusion
Antiplatelets Control
Relative risk
(webtable). The trials were done in 33 countries over six continents, and published between 1985 and 2005. Of these, we were unable to trace the investigators for seven trials, one trialist refused to participate, data were confi rmed as lost or unretrievable for 17 trials, and although available, were not supplied for two small trials. Ultimately, data were therefore available from 36 trials and 34 288 women (90% of randomised women). This paper presents the results from the 31 trials that recruited women in a primary prevention setting. These trials included 32 217 women and their 32 819 babies.10–14,25–50 Depending on the outcome, the minimum and maximum numbers of trials and women or babies available for individual analyses were between nine and 26 trials and between 7413 and 30 822 women or their babies. Aspirin was given alone in 27 trials, in a dose ranging from 50 to 150 mg per day, accounting for 98% of women in the dataset (n=31 678).10–14,25,29–36,38–40,43–50 Aspirin was given in combination with dipyridamole in three trials (177 women; two of these trials were three arm trials: Test for heterogeneity: χ²=31·19, df=23 (p=0·12), I²=26·3% aspirin alone vs aspirin and dipyridamole vs control).33,37,43 Test for overall effect: Z=2·86 (p=0·004) 362 women in three trials received other antiplatelet agents only (dipyridamole and/or heparin, ozagrel).
27 trials,10–14,25–29,31–36,38–42,44–49 including 97% of women, were
done in countries with a low perinatal mortality rate.51 Figure 2: Maternal pre-eclampsia (ordered by eff ect size)
Randomisation and therapy began before 20 weeks’ Fixed eff ect model used to calculate relative risks.
gestation in 59% of the women enrolled.
Of the 32 217 women who were recruited in a primary Number of Number of events (%)
Relative risk
prevention setting, 54% (n=17 544) were in their fi rst pregnancy, 92% (29 642) had a singleton pregnancy, 70% (22 657) were aged 20–35 years, and 90% (29 068) had at least one risk factor (which could include primiparity). Overall, 8% (2599) of these women developed Antiplatelet agents were associated with a signifi cant 10% reduction in the relative risk of both pre-eclampsia (p=0·004) and preterm birth before 34 weeks’ gestation (p=0·011) compared with control (fi gure 1 and fi gure 2). The data indicated a 10% reduction in the relative risk of the baby being small for gestational age and a *Using PARIS defi nition of blood loss ≥500 mL if supplied or trialists defi nition, but excluding two trials (Sibai et al51 9% reduction in the relative risk of stillbirth or baby death and Caritis et al53) due to data discrepancies with blood loss data. before discharge, although the 95% CI for both crossed Table 1: Other maternal outcomes
the point of no eff ect (fi gure 1). Overall, there was a www.thelancet.com Vol 369 May 26, 2007
Articles
signifi cant 10% reduction in the relative risk of our Number Number of events (%)
Relative risk (95% CI)
of trials
prespecifi ed composite outcome of pregnancy with any serious adverse outcome (any of the four main outcomes or maternal death; p=0·001). These data suggest that, in this population, for every 51 women treated with antiplatelet agents, a serious adverse outcome will be prevented in one pregnancy, and 114 women would need to be treated to prevent one case of pre-eclampsia. Results for maternal outcomes are shown in table 1. NICU=neonatal intensive care unit. SCU=special care unit. *Signifi cant diff erence in relative risk at p=0·05 level. There were no signifi cant diff erences between the two groups for any of these outcomes. Importantly, potential Table 2: Other infant outcomes
adverse eff ects—eg, ante-partum haemorrhage, placental abruption, and post-partum haemorrhage— Subgroup
Category
Antiplatelets Control
Relative risk
Interaction
were not signifi cantly diff erent between the two treatment groups. Additional baby outcomes are shown Pregnancy and medical history
in table 2. There was a 7% reduction in the relative risk of preterm birth before 37 weeks (p=0·003). Similarly, the data indicate a 13% reduction in the relative risk of preterm birth before 28 weeks, although the 95% CI cross the point of no eff ect. The reduction in relative risk of admission to a special care baby unit or neonatal intensive care unit associated with antiplatelet use rather than use of controls was small (4%) and not signifi cant. Although data were available for only nine trials (7413 babies), we found a 21% reduction in the likelihood of the baby receiving assisted ventilation (p=0·010), equivalent to an absolute reduction in risk of 1·3%, meaning that in this population, on average, 78 infants would need to be treated with antiplatelet agents (via their mothers) to prevent one from needing For the main outcome of pre-eclampsia, there was no evidence that women in any one of our prespecifi ed subgroups benefi ted more or less from the use of antiplatelet agents than those in any other subgroup (table 3). There was no evidence that using more than 75 mg of aspirin had more or less eff ect than a lower Current pregnancy
dose, or that commencing treatment before 20 weeks’ gestation was more or less benefi cial than starting later 924/10 935 1038/10 777 0·87 (0·80–0·95) in pregnancy (table 3). Nonetheless, since the absolute benefi t derived from antiplatelet agents also depends 1114/14 325 1206/14 187 0·91 (0·84–0·98) 0·67 on the woman’s underlying risk, the absolute eff ects and number needed to treat will vary by risk profi le Trial factors
We did the same subgroup analyses for the other four primary outcomes of baby death, preterm birth, small for gestational age infant, and pregnancy with any serious 1065/12 766 1163/12 784 0·92 (0·85–0·99) 0·23 adverse outcome. We found four subgroups with a signifi cant test for interaction (baby death: second or subsequent pregnancy with or without history of HDP=hypertensive disorder of pregnancy event including gestational hypertension, pre-eclampsia or eclampsia. There hypertensive disorders of pregnancy [p=0·007]; preterm were insuffi cient data to analyse fi rst pregnancy with family history of HDP (only 20 women with this event). High risk birth less than 34 weeks: second or subsequent pregnancy factor=current pregnancy with any of the following: autoimmune disease, renal disease, chronic hypertension, with or without history of hypertensive disorders of diabetes, abnormal uterine artery doppler fl ow studies, family history of HDP, multiple pregnancy, or an unspecifi ed pregnancy [p=0·012]; small for gestational age infant: risk factor; OR history of any of the following: gestational hypertension, pre-eclampsia, eclampsia, fetal or neonatal death. *Aspirin only trials. second or subsequent pregnancy with or without any high risk factor [p=0·032]; pregnancy with serious Table 3: Subgroup analyses of pre-eclampsia outcome
adverse outcome: single/multiple preg nancy [p=0·046]). www.thelancet.com Vol 369 May 26, 2007
Articles
Only the fi rst of these was signifi cant at the 1% level, Sample baseline PARIS relative risk
Number needed-to-treat
suggesting a potential greater benefi t in women who had event rate
a history of hypertensive disorders of pregnancy.
Overall results were similar when analyses were restricted to placebo controlled trials (data not shown), to trials of aspirin alone as the active treatment, and when extended to include trials with less than 80% of data available for a particular outcome (data not shown). Results were also unchanged when analyses of baby outcomes were repeated with numbers of women experiencing events, rather than number of babies experiencing events (data not shown).
For the 26 trials where data were available to allow comparison (webappendix), pooled results did not diff er substantially when the trialists’ own defi nition of pre-eclampsia was used (relative risk 0·88, 95% CI 0·81–0·96), compared with the pre-specifi ed PARIS defi nition19 (0·90, 0·83–0·97). Both are also consistent with our main analysis of all trials based on the best available defi nition: recoded to PARIS defi nition or using Table 4: PARIS number needed-to-treat with sample baseline event rates
trialists’ defi nition if recoding was not possible (0·90, 0·84–0·97). (table 2). Women in this small subgroup seemed to have Discussion
a larger than average reduction in the relative risk of a Our results show that antiplatelet agents produce stillbirth or baby death before discharge, as well as a moderate but consistent reductions in the relative risk of possible reduction in the risk of preterm birth when pre-eclampsia, preterm birth before 34 weeks’ gestation, treated with antiplatelet agents rather than control. and having a pregnancy with serious adverse outcome. Although prespecifi ed, these subgroup analyses should, There is no clear evidence that these agents are any more of course, be interpreted cautiously. As always when or less eff ective in reducing the relative risk for any there are multiple analyses, there is a serious risk of particular subgroup of women. The eff ect of antiplatelet being misled by the play of chance. We note that we did agents on pre-eclampsia seen here was much the same not fi nd similarly signifi cant interactions for pre-as that in the largest individual trial (7974 primary eclampsia, small for gestational age infant, or pregnancy prevention women, relative risk 0·88, 95% CI with serious adverse outcome in women with a history of 0·75–1·03).11,52 hypertensive disorder of pregnancy. Importantly, we Extensive subgroup and sensitivity analyses found no found no groups of women for whom there is evidence clear evidence that antiplatelet agents are more or less to justify withholding antiplatelet therapy. eff ective in preventing the development of pre-eclampsia One of the early concerns about the use of antiplatelet for any particular group of women. However, analyses of agents during pregnancy was the possibility of an high-risk categories were based on small numbers of increase in bleeding problems for either the woman or women, refl ecting the pattern of recruitment to the her child. This concern has been allayed by results from original trials, in which most women were at low to trials, including two that reported follow-up of the moderate risk of developing pre-eclampsia. For example, children at around 2 years of age,15 and the results of a few women had pre-existing renal disease or diabetes. As case-control study that indicate that aspirin use in early a result, our analysis was limited in its power to estimate pregnancy does not result in an increased risk of eff ects within these high-risk groups and to detect congenital abnormalities in infants.53 Our analyses diff erences, if any exist, between those with and without showed no change in the risk of post-partum or specifi c risk factors. Thus, despite having gathered an ante-partum haemorrhage between women who received extremely large dataset, the evidence base for particular antiplatelet agents and those who did not, nor was there groups of high-risk women remains limited and the most an eff ect on infant bleeding (table 1). Our analyses appropriate estimate of relative risk reduction remains highlight the problem of measuring and defi ning the overall estimate of 10%. post-partum haemorrhage. Two trials were excluded There was some suggestion that women in their second from the analysis of this outcome because of discrepancies or subsequent pregnancy with a history of a hypertensive between the data supplied for the dichotomous defi nition disorder of pregnancy might derive a larger benefi t from of post-partum haemorrhage, and for the estimated blood the use of antiplatelet agents than those in their second loss. Also, exploratory analyses found the overall results or subsequent pregnancy who did not have such a history to be sensitive to even small changes in the way we www.thelancet.com Vol 369 May 26, 2007
Articles
defi ned post-partum haemorrhage calculated from gathered in such a way that sharing for future individual estimated blood loss (eg, using greater than or equal to patient data meta-analyses is facilitated.
rather than greater than 500 mL; data not shown). Despite exhaustive eff orts to obtain individual patient Changes in defi nition had up to a four-fold eff ect on the data from all eligible trials, several of the early, small, estimated eff ect size, and infl uenced statistical positive trials were unable to retrieve their raw data. signifi cance. Data presented here are based on our pre- Also, fewer small negative trials are included in this specifi ed defi nition. Given the well-known diffi culties of review than might be expected.52 Although the inclusion accurately estimating blood loss at delivery, this outcome of such negative studies would lead to more conservative should be interpreted cautiously.
estimates, the numbers involved are so small that The defi nition of pre-eclampsia has long been overall estimates would be unlikely to change. controversial,22,23 and it has been argued that diff erences Publication bias is one possible explanation for the lack in the relative risk of pre-eclampsia reported in diff erent of small negative trials,59 but another is a diff erent case trials might merely result from the diff erent defi nitions mix in small trials compared with large trials. Because of the condition. This argument is also used as a criticism we have failed to confi rm any clear diff erences in the of meta-analyses based on aggregate data. An advantage eff ects of antiplatelet agents based on the characteristics of using individual patient data was that we were able to of individual women, the true explanation for the lack do prespecifi ed sensitivity analyses based on alternative of small negative trials might be that they remain defi nitions of pre-eclampsia. For some studies, the unpublished and inaccessible. This issue will be incidence of pre-eclampsia varied considerably explored further in future analyses. depending on whether the trialists’ or the PARIS Our data show that antiplatelet agents produce defi nition was used. For example, the CLASP and ECPPA moderate but consistent reductions in pre-eclampsia trials required a minimum rise in diastolic blood and its consequences, but there is no clear evidence that pressure in addition to minimum blood pressure values. such agents are any more or less eff ective in reducing The PARIS defi nition did not require this minimum rise the relative risk for any particular subgroup. This because it is no longer included in most international information should be discussed with women at risk of classifi cations.22,24,54–56 The event rate for pre-eclampsia in pre-eclampsia to help them make informed choices these two trials in the PARIS analysis is therefore higher about their antenatal care. Whether individual women than that in the original trial reports, although the will choose to take antiplatelet agents might depend on relative risks did not alter substantially. Despite an assessment of their absolute risk. From a public-health diff erences for individual trials, the overall results across perspective, especially for populations with a high risk all trials did not change substantially when diff erent of pre-eclampsia, even these moderate benefi ts could defi nitions were used. make more widespread use of antiplatelet agents Although well established in cancer and cardiovascular worthwhile. medicine, meta-analyses of individual patient data have Contributors
rarely been used in other areas of health care.57 Advantages LMA participated in protocol development, data collection, data analysis, of such an approach include the ability to do extensive and interpretation. LD, DJH-S, and LAS participated in protocol data checking to ensure the quality of the dataset. The development, data analysis and interpretation. All members of the writing committee saw and approved the fi nal version. All active PARIS approach could also circumvent the many potential collaborators were sent the paper as prepared for submission and given biases associated with publication and published data.58 the opportunity to comment on the draft manuscript.
The ability to standardise analyses also improves the PARIS Collaborative Group
robustness of fi ndings, as does the potential to analyse Writing committee: Lisa Askie, Lelia Duley, David Henderson-Smart, data for a more complete set of outcomes than from the Lesley Stewart.
published literature. In our analysis, the trialists provided Steering group: Lelia Duley, David Henderson-Smart, Lisa Askie, Marian Showell, Lesley Stewart, Barbara Farrell, Mike Clarke, data for outcomes that have not been consistently James King, Christine Roberts. reported in trial reports or other systematic reviews—eg, Active PARIS collaborators (data supplied or data confi rmed as assisted ventilation and post-partum haemorrhage. lost/unavailable), by trial: Australia 1988: B Trudinger, C Cook; Collecting individual patient data also permits subgroup Australia 1993: B Walters, C Michael; Australia 1995: J Newnham, J Francis; Australia 1995a: R North, P Kincaid-Smith; Australia 1996: analyses that are generally impossible or limited if J Morris, C Cook; Australia 1996a: R North, C Ferrier; Australia 1997: attempted with aggregate published data. Another E Gallery, M Ross; Austria 1992: H Schröcksnadel; BLASP 1998: advantage, specifi c to this fi eld, is the ability to link K Cruikshank, A Hennis; China 1996: Z Wang, H Wang; China 1999: mother and baby outcomes. Furthermore, our dataset M Rogers, H Fung; CLASP 1994: C Redman, M de Swiet; Colorado 1993: R Porreco, D Hickok; ECPPA 1996: A Atallah, B Farrell; EPREDA 1991: will enable risk modelling, further investigation of the M Beaufi ls, S Uzan; ERASME 2003: D Subtil, F Maillard; Finland 1993: eff ect of diff erent antiplatelet load, and the investigation L Viinikka, O Ylikorkala; Finland 1997: P Zimmerman, E Kujansuu; of the eff ect of this therapy for women with gestational Finland 1997a: M Tulppala, O Ylikorkala; Finland 2002: M Vainio, hypertension. The individual patient data approach thus J Maenpaa; France 1985: M Beaufi ls, S Uzan; France 1990: R Azar, D Turpin; Germany 2000: D Grab, M Erdmann; India 1991: V Grover, off ers considerable potential in the perinatal fi eld. Those S Sachdev; India 1994: U Roy, S Pan; Iran 2002: A Taherian, A Shirvani; planning and doing trials should ensure that data are Israel 1989 and Israel 1990: E Schiff ; Italy 1989: T Frusca, A Benigni; www.thelancet.com Vol 369 May 26, 2007
Articles
Italy 1993: F Parazzini, E Ricci; Italy 2004: F Chiaff arino, F Parazzini; 11 CLASP (Collaborative Low-dose Aspirin Study in Pregnancy) Jamaica 1998: A McCaw-Binns, J Golding; Japan 1999: H Seki, S Takeda; Collaborative Group. CLASP: a randomised trial of low-dose aspirin Netherlands 1986, Netherlands 1989, and Netherlands 1991: G Dekker, for the prevention and treatment of pre-eclampsia among H Wallenburg; New Zealand 1999: L McCowan, J Harding; 9364 pregnant women. Lancet 1994; 343: 619–29.
New Zealand 2000: L Chamley, N Pattison; PERGAR 1989: S Uzan, 12 Golding J. A randomised trial of low dose aspirin for primiparae in F Maillard; Russia 1994: I Sidorova, V Rogov; South Africa 1988: pregnancy. BJOG 1998; 105: 293–99.
A Railton, J Anthony; Spain 1997 and Spain 2003: R Hermida, D Ayala; 13 Rotchell YE, Cruikshank JK, Phillips GM, et al. Barbados Low Dose Spain and other 2000: L Cabero; Tanzania 1995: C Ramaiya, Aspirin Study in Pregnancy (BLASP): a randomised trial for the R Mpembeni; Thailand 1996: Y Herabutya, A Thakkinstian; UK 1990: prevention of pre-eclampsia and its complications. BJOG 1998; P McParland, G Chamberlain; UK 1992: F Broughton Pipkin, K Louden; 105: 286–92.
UK 1992b and UK 1995: R Farquharson; UK 2003: C Yu, 14 Caritis S, Sibai BM, Hauth J, et al. Low-dose aspirin to prevent A Papageorghiou; USA 1993: J Hauth, S Cliver; USA 1993a: B Sibai, preeclampsia in women at high risk. N Engl J Med 1998; 338: 701–05.
E Thom; USA 1994: P August, G Helseth; USA 1998: S Caritis, E Thom; 15 Duley L, Henderson-Smart DJ, Knight M, King J. Antiplatelet Venezuela 2000: C Rivas-Echeverria, L Molina; Zimbabwe 1998: agents for preventing pre-eclampsia and its complications. Cochrane Database Syst Rev 2003; 1: CD004659.
R Byaruhanga, T Chipato. Analysis support and data managers for the overall project: Jayne Tierney, 16 Lindheimer MD. Unraveling the mysteries of preeclampsia. Am J Obstet Gynecol 2005; 193: 3–4.
17 Jeyabalan A, Caritis S. Antioxidants and the prevention of Analysis support and data managers for specifi c trials: Sarah Ayers, preeclampsia—unresolved issues. N Engl J Med 2006; 354: 1841–43.
18 Lindheimer MD, Sibai BM. Antioxidant supplementation in Confl ict of interest statement
pre-eclampsia. Lancet 2006; 367: 1119–20.
We declare that we have no confl ict of interest.
19 The Perinatal Antiplatelet Review of International Studies (PARIS) Collaboration Steering Group on behalf of the PARIS Collaboration. Acknowledgments
Antiplatelet agents for prevention of pre-eclampsia and its The PARIS Collaboration primary funder was the National Health and consequences: a systematic review and individual patient data Medical Research Council (NHMRC) of Australia, through a 3-year project meta-analysis. BMC Preg Childbirth 2005; 5: 7.
grant (ID 253636) for the overall project administration base in Sydney, 20 Editorial Team: Cochrane Pregnancy and Childbirth Group, About and a Sidney Sax Public Health Postdoctoral Fellowship (ID 245521) for the Cochrane Collaboration (Cochrane Review Groups (CRGs)). LMA, based in Oxford and Sydney. Additional support was provided by the Resource Centre for Randomised Trials and the UK Cochrane Centre 21 Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring (Oxford, UK); the Medical Research Council Clinical Trials Unit (London, inconsistency in meta-analyses. BMJ 2003; 327: 557–60.
UK); and the NHMRC Clinical Trials Centre (University of Sydney, 22 Brown MA, Hague WM, Higgins J, et al. The detection, Australia). Many thanks go to all the women who were involved in the investigation and management of hypertension in pregnancy: full numerous trials included in the PARIS Collaboration. Special thanks are consensus statement. Aust N Z J Obstet Gynaecol 2000; 40: 139–55.
extended to Lynn Hampson and Sonja Henderson for help with the search 23 Higgins J, De Swiet M. Blood-pressure measurement and strategy and identifying trial reports. Thanks also to many others who classifi cation in pregnancy. Lancet 2001; 357: 131–35.
assisted the project in a variety of ways including administrative assistance, 24 Roberts JM, Pearson GD, Cutler JA, Lindheimer MD. Summary of translation, and methodological and statistical advice: Charles Algert, the NHLBI Working Group on research on hypertension during Doug Altman, Gerard Breart, Peter Brocklehurst, Iain Chalmers, pregnancy. Hypertens Pregnancy 2003; 22: 109–27.
M S Elmahaishi, Joanna Emsley, Nicola Flack, Marc Keirse, Marian Knight, 25 Michael CA, Walters BNJ. Low-dose aspirin in the prevention of Shireen Meher, Mohsen Mirzaie, Jim Neilson, Kypros Nicolaides, pre-eclampsia: current evaluation. In: Teoh ES, Shan Ratnam S, Barbara Roberts, Jim Roberts, Jeff rey Robinson, Lara Rydzewska, Macnaughton M, eds. Maternal physiology and pathology. The Nelson Sass, Emma Smith, Karla Soares-Weiser, Patsy Spark, Cathy Spong, current status of gynaecology and obstetrics series, volume 4. Charlene Thornton, Sylvaine Verhulst, and Silke Walleser.
Carnforth: Parthenon Pub Group Ltd, 1993: 183–89.
26 Kincaid-Smith P, North RA, Fairly KF, Kloss M, Ihle BU. References
Prevention of pre-eclampsia in women with renal disease: a World Health Organization International Collaborative Study of prospective randomized trial of heparin and dipyridamole. Hypertensive Disorders of Pregnancy. Geographic variation in the Nephrology 1995; 1: 297–300.
incidence of hypertension in pregnancy. Am J Obstet Gynecol 1988; 27 Morris JM, Fay RA, Ellwood DA, Cook C, Devonald KJ. A 158: 80–83.
randomized controlled trial of aspirin in patients with abnormal Khan KS, Wojdyla D, Say L, Gulmezoglu AM, Van Look PF. WHO uterine artery blood fl ow. Obstet Gynecol 1996; 87: 74–78.
analysis of causes of maternal death: a systematic review. Lancet 28 Ferrier C, North R, Kincaid-Smith P. Low dose aspirin delays the 2006; 367: 1066–74.
onset of pre-eclampsia in pregnancies with abnormal uteroplacental Rosenfi eld A, Maine D. Maternal mortality—a neglected tragedy. circulation. Proceedings of the 10th World Congress of the ISSHP, Where is the M in MCH? Lancet 1985; 2: 83–85.
Seattle, Washington, USA. Aug 4–8, 1996. 151.
Mahler H. The safe motherhood initiative: a call to action. Lancet 29 Gallery ED, Ross MR, Hawkins M, Leslie GI, Gyory AZ. Low-dose 1987; 1: 668–70.
aspirin in high risk pregnancy? Hypertens Preg 1997; 16: 229–38.
Redman CW, Sargent IL. Latest advances in understanding 30 Wang ZH, Li WJ. A prospective randomized placebo-controlled trial pre-eclampsia. Science 2005; 308: 1592–94.
of low dose aspirin for prevention of intra uterine growth Redman CW, Bonnar J, Beilin L. Early platelet consumption in retardation. Chin Med J 1996; 109: 238–42.
pre-eclampsia. BMJ 1978; 1: 467–69.
31 Rogers MS, Fung HY, Hung CY. Calcium and low-dose aspirin Janes SL, Kyle PM, Redman C, Goodall AH. Flow cytometric prophylaxis in women at high risk of pregnancy-induced detection of activated platelets in pregnant women prior to the hypertension. Hyperten Pregnancy 1999; 18: 165–72.
development of pre-eclampsia. Thromb Haemost 1995; 74:
32 ECPPA (Estudo Colorativo para Prevencao da Pre-eclampsia com Aspirina) Collaborative Group. ECPPA: randomised trial of low Bussolino F, Benedetto C, Massobrio M, Camussi G. Maternal dose aspirin for the prevention of maternal and fetal complications vascular prostacyclin activity in pre-eclampsia. Lancet 1980; 2: 702.
in high risk pregnancies. BJOG 1996; 103: 39–47.
Masotti G, Galanti G, Poggesi L, Abbate R, Neri Serneri GG. 33 Uzan S, Beaufi ls M, Breart G, Bazin B, Capitant C, Paris J. Diff erential inhibition of prostacyclin production and platelet Prevention of fetal growth retardation with low-dose aspirin: aggregation by aspirin. Lancet 1979; 2: 1213–17.
fi ndings of the EPREDA trial. Lancet 1991; 337: 1427–31.
10 Sibai BM, Caritis S, Thom E, et al. Prevention of preeclmapsia with 34 ERASME collaborative group. Asprin (100 mg) used for prevention low-dose aspirin in healthy, nulliparous pregnant women. of pre-eclampsia in nulliparous women: the Essai Regional Asprine N Engl J Med 1993; 329: 1213–18.
Mere-Enfant study (Part 1). BJOG 2003; 110: 475–84.
www.thelancet.com Vol 369 May 26, 2007
Articles
35 Zimmerman P, Eirio V, Koskinen J, et al. Eff ect of low dose aspirin 47 Hauth J, Goldenberg R, Philips J, Copper R, DuBard M, Cutter G. treatment on vascular resistance in the uterine, uteroplacental, Low-dose aspirin therapy to prevent pre-eclampsia: safety renal and umbilical arteries. Eur J Ultrasound 1997; 5: 17–30.
considerations. Am J Obstet Gynecol 1993; 168: 389.
36 Vainio M, Kujansuu E, Iso-Mustajarvi M, Maenpaa J. Low dose 48 August P, Helseth G, Edersheim TG, et al. Sustained release, acetylsalicylic acid in prevention of pregnancy-induced low-dose aspirin ameliorates but does not prevent preeclampsia hypertension and intrauterine growth retardation in women with (PE) in a high risk population. Proceedings of 9th International bilateral uterine artery notches. BJOG 2002; 109: 161–67.
Congress, ISSHP; Sydney, Australia: 1994, 72.
37 Grover V, Shabnam S, Kumari S. Evaluation of dipyridamole & 49 Rivas-Echeverria CA, Echeverria Y, Molina L, Novoa D. Synergic use aspirin in prevention and management of intrauterine growth of aspirin, fi sh oil and vitamins C and E for the prevention of retardation. J Perinat Med 1991; 19: 104.
pre-eclampsia. Hypertens Preg 2000; 19: 30.
38 Schiff E, Peleg E, Goldenberg M, et al. The use of aspirin to prevent 50 Byaruhanga RN, Chipato T, Rusakaniko S. A randomized controlled pregnancy-induced hypertension and lower the ratio of trial of low-dose aspirin in women at risk from pre-eclampsia. thromboxane A2 to prostacyclin in relatively high risk pregnancies. Int J Gynecol Obstet 1998; 60: 129–35.
N Engl J Med 1989; 321: 351–56.
51 WHO. World health report 2005: make every mother and child 39 Italian Study of Aspirin in Pregnancy. Low-dose aspirin in count. Geneva: World Health Organization, 2005.
prevention and treatment of intrauterine growth retardation and 52 Duley L, Henderson-Smart DJ, Meher S, King J. Antiplatelet agents pregnancy-induced hypertension. Lancet 1993; 341: 396–400.
for preventing pre-eclampsia and its complications. 40 Chiaff arino F, Parazzini F, Paladini D, et al. A small randomised Cochrane Database Syst Rev 2007; 2: CD004659.
trial of low dose aspirin in women at high risk of pre-eclampsia. 53 Nørgård B, Puhó E, Czeizel AE, Skriver MV, Sørensen HT. Eur J Obstet Gynecol Reprod Biol 2004; 112: 142–44.
Aspirin use during early pregnancy and the risk of congenital 41 Seki H, Kuromaki K, Takeda S, Kinoshita K, Saton K. Trial of abnormalities: a population-based case-control study. prophylactic administration of TXA2 synthetase inhibitor, ozagrel Am J Obstet Gynecol 2005; 192: 922–23.
hydrochloride, for pre-eclampsia. Hypertens Preg 1999; 18: 157–64.
54 Davey DA, MacGillivray I. The classifi cation and defi nition of the 42 Uzan S, Beaufi ls M, Bazin B, Danays T. Idiopathic recurrant fetal hypertensive disorders of pregnancy. Am J Obstet Gynecol 1988; 158:
growth retardation and aspirin-dipyridamole therapy. Am J Obstet Gynecol 1989; 160: 763–64.
55 Milne F, Redman CW, Walker J, et al. The pre-eclampsia 43 Railton A, Davey DA. Aspirin and dypyridamole in the prevention community guideline (PRECOG): how to screen for and detect of pre-eclampsia: eff ect on plasma prostanoids 6 keto PG1a and onset of pre-eclampsia in the community. BMJ 2005; 330: 576–50.
TXB2 and clinical outcome of pregnancy. Proceedings of 6th World 56 1999 World Health Organization-International Society of Congress of the ISSHP; Toronto, Canada; 1988, 60.
Hypertension Guidelines for the Management of Hypertension. 44 Hermida RC, Ayala DE, Iglesias M. Administration time-dependent Guidelines Subcommittee. J Hypertens 1999; 17: 151–83.
eff ects of aspirin on blood pressure in pregnant women. 57 Scott JR. Immunotherapy for recurrent spontaneous abortion: Hypertension 2003; 2: 651–56.
analysis 2. Am J Reprod Immunol 1994; 32: 279–80.
45 Louden KA, Broughton Pipkin F, Symonds EM, et al. A randomised 58 Stewart LA, Tierney J, Burdett S. Do systematic reviews based on placebo-controlled study of the eff ect of low dose aspirin on platelet individual patient data off er a means of circumventing biases reactivity and serum thromboxane B2 production. BJOG 1992; 99:
associated with trial publications? In: Rothstein HR, Sutton AJ, Borenstein M, eds. Publication bias in meta-analysis: prevention, 46 Yu CK, Papageorghiou AT, Parra M, Palma D, Nicolaides KH. assessment and adjustments. Chichester, England; Hoboken, NJ: Randomised controlled trial using low dose aspirin in the J John Wiley & Sons Ltd; 2005: 261–86. prevention of pre-eclampsia in women with abnormal uterine artery 59 Dickersin K. The existence of publication bias and risk factors for Doppler at 23 weeks gestation. Ultrasound Obstet Gynecol 2003; 22:
its occurrence. JAMA 1990; 263: 1386–89.
www.thelancet.com Vol 369 May 26, 2007

Source: http://www.medi-gate.jp/contents/client/1229/1229-top-pdf1.pdf